I stopped too, after they threw back some of my reports because they were missing some completely unrelated generic diagnostic info (like which MacBook generation was I using, even if the bug was 100% reproducible everywhere), even though I’ve provided them a minimal working example on GitHub, and even the method name of the problematic private API.
Step 1: *systematically kill the middle class over a generation*
Step 2: "noooooo, why are these workers not pushing themselves like they used to? are you not believing in our mission statement? but we've introduced Pride Month this year! what else do you want???"
And you're absolutely right, despite the downvotes, but here's the coldest of all hot takes: it doesn't matter if it's shit. It just has to be there.
Doomscrolling and modern web products were never about quality, but rather about giving the user the hope of potentially seeing something of quality.
Which means that my cousin magic_hamster who said "this can just be an algorithm" is also right [1]. In most cases there's no need to leverage AI here. 90% of everything is gonna be crap, and people historically often get addicted to low-quality crap anyway.